I suppose that's technically true, but what I mean is that injecting e.g. water would not cause cell death. Yes if you attach yourself to a IV water pump then you can kill yourself by totally de-balancing your body chemistry, but water is not itself a toxin. It doesn't cause cell death or attack the body in any way.
By the way, the fact that the body can flush it out doesn't mean it's not a toxin. Alcohol is gone from the body within hours of getting drunk but it's still a toxin, technically speaking.
In contrast, the mRNA vaccine works by getting cells to do things that cause them to be destroyed by the immune system ASAP. Even small amounts are thus (cyto)toxic and the question is how much can the body tolerate.
One of the things that makes me uneasy about the mRNA tech is there seems to be very little discussion or research about dosages, and in particular how cell death levels compare between getting vaccinated and actually being infected with the real thing. This would seem to be fundamental because at some level, the vaccine is actually creating the same problem inside the body that the virus is, the only difference is that the virus can self replicate whereas the mRNA does not (except apparently in the case of so-called "self-amplifying mRNA" vaccines, see below). Thus the question of dosage is critical. If being infected with the virus causes 10x less cell death than the vaccine then I want to take my chances with the virus. If the vaccine causes 10x less cell death than a real viral infection, then it looks better. As ultimately, it's cell death and the fight to destroy those cells that makes people sick.
Such numbers are (nearly?) impossible to find. Vaccine dosages are very different between the different manufacturers. The Moderna vaccine has a dose more than 3x higher than the Pfizer vaccine, for example (100 µg vs 30 µg), which is itself 3x higher than the dose for the Curevac vaccine (12µg).[1] It's a bit unclear how these very different numbers were arrived at or what they're being calibrated against.
W.R.T. the SAM vaccines, also in [1] we find the following text:
"The other type of mRNA vaccines, SAMs, do not only encode the target antigen but also RNA polymerase encoding ‘self-amplifying’ factors derived from Alphavirus ... These are in turn transcribed to many coding mRNA molecules, leading to prolonged and enhanced antigen expression ... The two SAM vaccines in clinical development are nCoVsaRNA by the Imperial College London and ARCT-021 by Arcturus/Duke-NUS (both as mRNA-LNP)."
It's very unclear to me, as a layman, how this artificial "self amplifying RNA" that incorporates elements from a virus is so very different from an actual virus, given that the SARS-CoV-2 is itself basically RNA surrounded by a coating in a form that can self-replicate. Traditional vaccines are virus based but the virus is inert. The new vaccines appear to be ending up at the same end-point but without the whole making it inert part.
It'd got to be higher than that. In Mexico, population 127 mil, excess deaths are about 622k which is 0.49% of the population dead.
The number comes from meta-studies that examine a large range of IFR studies. There is a lot of variance between them - they don't all arrive at the same value, presumably because IFR is inherently difficult to measure. Infected people who don't feel sick enough to report are the dark matter of epidemiology.
However, excess death values are quite tricky to evaluate because they're excess relative to some model of what should have happened in an alternate timeline where the event in question never occurred. For example you can calculate an excess death value for Sweden of zero by making one or two plausible assumptions, or you can calculate a higher value using equally plausible assumptions, and who is to say which is correct? Ultimately you can't know what would have happened otherwise, only extrapolate from the past and try to guess. Some of those extrapolations look suspect on examination (e.g. using fixed averages, or very short timespans), just like everything else around this topic.
[1] https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8032477/